Controlling Democracy: The PrincipalAgent Problems

in Election Administration
R. Michael Alvarez and Thad E. Hall
Election reform has become a major issue since the 2000 election, but little consideration has been
given to the issues associated with managing them. In this article, we use principalagent theory to
examine the problems associated with Election Day polling place voting. We note that Election Day
voting manifests problems that agency theory shows are difficult to overcome, including adverse
selection of and shirking by poll workers. We then examine alternate methods of voting, such as early,
absentee, and Internet voting, and show how these reforms can mitigate many of the more severe
principalagent problems in election management.
KEY WORDS: election reform, public management, principalagent theory
Few citizens gave much consideration to the issue of election administration
in the United States before November 7, 2000. However, scholars have recognized
that election administration has been a problem for some time. In 1934, the
Brookings Institution issued a report stating, There is probably no other phase of
public Administration in the United States, which is so badly managed as the
conduct of elections (Harris, 1934, p. 1). The irony here is rich. Elections are where
the public makes primary decisions that affect all citizens and all administrators.
It is where public preferences manifest themselves in decisions about who will run
all levels of government and, through the initiative and referenda process, even
how the government will be run. Yet the history of election administration is one
where frequently ill-equipped, poorly trained, part-time administrators have been
trusted with managing this critical democratic function.
1
After the 2000 presidential election, the current debate over election reform
began, with a primary focus on voting equipment and on developing procedural
remedies for aws observed in the current election process. Consider, for example,
the election reform legislation enacted by Congressthe Help America Vote Act
of 2002 (Public Law 107252). This legislation was primarily designed to provide
states with money to purchase new voting equipment and to encourage states to
implement a series of specic procedures, such as statewide voter registration
systems, voter education programs, and provisional voting (Liebschutz &
Palazzolo, 2005; Alvarez & Hall, 2005).
The Policy Studies Journal, Vol. 34, No. 4, 2006
491
0190-292X 2006 The Policy Studies Journal
Published by Blackwell Publishing. Inc., 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington Road, Oxford, OX4 2DQ.
This law and many national and state-level reform efforts (e.g., Governors
Select Task Force on Elections Procedures, Standards and Technology, 2001;
National Commission on Federal Election Reform, 2002) typically focused on
changing the current election system without considering whether the model of
election administration that is used today is appropriate.
2
That is, why do most
people vote at polling places on a single day at a single location? Why are voting
places sometimes located in private homes and garages, or in private businesses?
Why do volunteers or poorly paid people staff these polling places? What are the
managerial complexities created by using such an election system? Are there other
models for elections that would strengthen the ability of election administrators to
control and manage the election process?
In this article, we use the logic of principal agency to analyze the problems that
arise in the existing election system built around polling place voting on a single
day. We then identify alternate means of serving voters that election administrators
can employ to retain greater control over the election process. Using this theo-
retical construct, we consider the various components of poll site voting that
administrators must manage in the current system and use cases from recent
elections to illustrate the difficulties associated with managing each component of
the system. We then explore how other means of serving voterssuch as vote
centers, vote-by-mail, early voting, and Internet votingcan allow election admin-
istrators to mitigate many of the serious agency problems associated with election
administration. We conclude with a discussion of how administrators can use
experimentation to test these administration techniques and present systematic
data from Oregon and the United Kingdom to bolster this point.
PrincipalAgent Theory
Principalagent theory has been used extensively in public administration,
implementation analysis, and political science to examine the problems associated
with management and administration in a decentralized environment.
3
As
Waterman and Meier (1998, p. 174) note, the principal agent model . . . is in
essence a theory about contractual relationships between buyers and sellers. A
critical management problem in most organizations is delegation. Delegation
occurs when a principalwho wants an activity accomplished but cannot easily
perform the taskinstead hires an agent to accomplish the task. Unfortunately,
just as principals cannot do the task themselves, they often have difficulty knowing
if they hired the right person and whether the task is being accomplished
appropriately.
The two problemshiring the right agent and knowing that they will do the
job appropriatelyare known respectively as adverse selection and moral hazard.
Moral hazards arise because the principal and agent often have conicting goals
and views of risk; the principal cannot assume the agent will act in the principals
best interest. This conict is exacerbated because monitoring the actions of an
agent can be costly. Adverse selection problems leave principals in the position of
not knowing if they have hired the right person for the job and if the agent is
492 Policy Studies Journal, 34:4
actually who the person she has represented.
4
Key aspects of the principalagent
relationship are the discretion available to the agent, the specic charge given to
the agent, and the formality of the agents contract (Mitnick, 1980; Waterman &
Meier, 1998).
Principalagent problems are inherently information related, but they are not
necessarily intransigent. As Thompson (1998) notes, corporations and governments
are typically able to nd satisfactory solutions to principalagent problems;
business and governments prosper even in the face of these problems. The four
strategies that are typically employed by principals to overcome this information
asymmetry are (i) contracting, (ii) candidate screening requirements, (iii) monitor-
ing requirements, and (iiii) institutional checks. However, not all administrative
structures and not all problems allow for these mitigation strategies to be easily
implemented. The administration of elections is an example of a policy area that
has complicated principalagent problems that are not easily resolved.
Designing contracts between principals and agents is complicated because
both sides have an incentive to ensure that the contract maximizes their own
position (e.g., Donahue, 1989, especially ch. 5; Kiewiet & McCubbins, 1991, ch. 2).
Cooper (2003) has shown that governments rarely have the resources necessary
to achieve their contractual goals. From the principals perspective, the contract
should constrain the agents ability to act contrary to the goals of the principal, and
maximize the amount of work that the agent will be completing. Designing
appropriate contracts is difficult, and even with one, it can be costly for the
principal to enforce the terms of the contract and to gather the necessary infor-
mation to ensure that the agent is fullling its end of the agreement.
A well-designed contract is of limited benet if the organization cannot
overcome adverse selection problems. Given the costs associated with collecting
meaningful performance data after an agent has been hired, it is important to hire
the right person at the outset. However, developing an effective screening process
can also be exceedingly difficult. Just as there is a lemon problem in the used-car
market, underqualied candidates apply for jobs that maximize their pay relative
to their skills (e.g., Akerlof, 1970). Principals must spend valuable resources to
pierce the veil of deception that agents have an incentive to cloak themselves
within.
The difficulties associated with contracting and the adverse selection problem
leads many principals to implement monitoring requirements, with performance
reports and the like being commonly used (Hall & OToole, 2000). There are many
techniques for determining if agents are acting appropriately (e.g., Banks, 1989;
Banks & Weingast, 1992; Bendor, Taylor, & van Gaalen, 1987), but each technique
is costly. Monitoring takes time and other resources away from achieving the goal
that is being monitored, and principals can end up ooded with information that
they must sort through to determine if agents are behaving appropriately. Also, it
can take a signicant amount of time for the principal to observe whether the
agent is actually acting consistently with job expectations.
Institutional mechanisms can also be enacted to limit the ability of agents to
act outside the interest of principals. The most common type of check is to limit
Alvarez/Hall: Controlling Democracy 493
the ability of the agent to act unilaterally. Requiring agents to receive sign-off
before acting and requiring multiple actors to agree on a single decision are
examples of this sort of institutional check. The goal behind these institutional
checks is to limit the damage that an agent can do if they decide to directly act
against the interests of the organization.
Our interest is in considering whether the problems identied through
principalagent theory can be easily mitigated in the context of American election
administration. We specically focus on four aspects of the principalagent model:
1. The principalagent dilemma can be eliminated if the principal chooses not to
delegate.
2. The methods available to principals for minimizing the effects of the principal
agent dilemmacontracting, monitoring, and the likerequire time to pass so
that the agent can act and these actions can be evaluated.
3. Overcoming the adverse selection problem requires a large pool of potential
agents from which to select, so the principal is in a position to reject all
unqualied agents.
4. The ability of the principal to negotiate successfully with agents is contingent
on the ability of the principal having resources with which to negotiate. A
contract, by denition, requires both sides to agree on terms. Any leverage that
the principal can have over the agent is likely to bring the negotiations to a
resolution.
With this principalagent framework in place, we can apply it to various models
of election implementation in order to determine the factors that alleviate or
exacerbate the ability of local election officials to control the election management
process. In the next section, we apply this framework to the traditional poll site
voting model and show how it creates a series of principalagent problems that
produce signicant difficulties for election administration. We then compare the
polling place model with alternate models that have fewer principalagent
concerns.
Applying PrincipalAgent Theory to the Current Election System
Election administration is a multistage process. First, there are the interactions
between the administrator and the candidates that set the stage for the election.
Second, there are interactions between administrators and various entities that are
involved in the ballot and voter information production process. For example,
election administrators need to design and proofread ballots, prepare voter guides
(in the limited number of localities that utilize them), and perform any required
ballot language translations (e.g., General Accounting Office [GAO], 1997). His-
torically, election administration has not faltered at these two points, in large
measure because of the control that election administrators have over the process.
5
However, as elections are becoming more complicatedespecially involving
494 Policy Studies Journal, 34:4
technology that puts the election official at the reliance of vendors to provide
support to electronic registration and voting systemsthere are more principal
agent problems arising at this point in the process. In the interest of parsimony, we
focus our analysis on the third part of the election process, the voter registration
and the voting process, where principalagent problems are especially acute. These
failures occur largely because of the agency problems produced by complicated
registration and voting systems, as well as the irregularity in the occurrence of
elections.
6
The problems associated with polling places are numerous and principal
agent theory suggests that they are more likely to be intransigent. The all-in-
one-day nature of poll site voting forces election officials as principals to rely
heavily on various agents for the conduct of voting. Local election officials (LEOs)
are at the mercy of the owners of poll sites, the poll workers, and the voting
equipment vendors to ensure that elections can be a success. Consider how the
GAO (2001a) described the task of election administration in preparing for Election
Day:
Although there was variation in how jurisdictions prepared for and
conducted the November 2000 election, behind the scenes, election admin-
istration officials across the United States performed similar duties. Before
election day, they designed ballots, marshaled and trained thousands of
workers to staff the polls on election day, located and prepared polling
places, organized and delivered voting equipment and supplies, and
educated citizens. On the day of the election, election officials shared control of
the election with an army of poll workers who staffed and oversaw the polls where
votes were cast and ballots collected. (GAO, 2001a, pp. 1516, emphasis
added).
This description of polling place voting explicitly notes the principalagent
nature of polling place voting and suggests that the high level of delegation is a
recipe for problems. Moreover, the GAO also found that both the adverse selection
and moral hazard aspects of the principalagent dilemma exist in elections, noting:
. . . 57 percent of voting jurisdictions nationwide encountered major prob-
lems in conducting the November 2000 election. Although all jurisdictions
did not experience the same problems, about half of all jurisdictions cited
problems with recruiting enough qualied poll workers. However, few
election jurisdictions systematically collected information on how well
their jurisdictions administered the election. (GAO, 2001a, pp. 1516).
Poll Workers
The GAO analysis identies several principalagent problems, especially
adverse selection problems, inherent in this system. The rst point of delegation
is to poll workers, who manage an election at poll sites. One study found that 51
percent of jurisdictions nationwide had signicant problems recruiting enough poll
Alvarez/Hall: Controlling Democracy 495
workers (GAO, 2001a, pp. 158160). A small pool of poll workers makes it difficult
for a LEO to screen poll workers adequately because they cannot afford to turn
many poll workers away. Recruiting is made all the more difficult because election
administrators do not have the nancial resources to provide signicant monetary
incentives to poll site workers, and because polling place workers are typically
required to work the entire election day (which can in many places be as long as
a 12-hour day). Payment for poll workers typically ranges between $50 and $150.
Local election officials instead rely on their locality having a large pool of
civic-minded citizenry who are willing to serve the community on Election Day
and who can nd the time away from work and family to do this. Election
administrators report that a disturbingly high number of recruited poll workers
fail to show up on Election Day, which implies that relying solely on civic
mindedness has its limits.
7
Additionally, as the size of the election jurisdiction increases, the ability of
election officials to screen workers declines precipitously. For example, in Alex-
andria, VA, where there are 24 poll sites and approximately two hundred poll
workers, election officials can at least screen the chief poll worker for each precinct
with some care. By contrast, there are 5,000 poll sites in Los Angeles County, CA,
with 5,000 chiefs and 25,000 poll workers. Screening this many individuals is
almost an impossible task, putting LEOs in the position of having to trust the
implementation of election services to individuals about whom they know little, if
anything.
This screening is important because of the autonomy poll workers have on
Election Day. In many jurisdictions, these individuals have the nal authority on
interpreting guidance in areas such as deciding who can vote and determining
voter intent (GAO, 2001a, p. 160). Even if these individuals do not legally have
the nal authority, poll workers are street-level bureaucrats and the election
officials are forced to rely on them to follow regulations and not act as nal arbiter
of disputes (Lipsky, 1980; Montjoy & Slaton, 2002). In the end, though, given the
nature of Election Day polling operations, poll workers end up with a vast degree
of discretionary authority, if they choose to use that authority, as there is little
oversight possible and few incentives available to keep poll workers from exer-
cising their discretion. In many states, poll workers count ballots and have nal
determination of voter intent, especially in the counting of absentee ballots. They
also exercise discretion in deciding who must show identication at the polls, how
to handle voters who cannot be readily located in the voter registration rolls, and
how to handle problems at the polls. This discretion has led at least one election
official to refer to poll workers as street level lawyers who make legal decisions
on the y on Election Day.
The following examples illustrate this point. In the 2005 mayoral runoff
election in the city of Los Angeles, our observing team visited a polling place in
the Eagle Rock area of northwestern Los Angeles.
8
There we observed polling
place practices in a church that is associated with a senior living facility. The head
poll worker in this voting location had taken upon herself to create and to post in
each voting station a sign that noted that voters could cast an assisted ballot only
496 Policy Studies Journal, 34:4
with the permission of the polling place workers. While this is one way to interpret
California election regulations on assisted voting, whether the polling place
workers were correct in exercising their discretion by posting signs in voting
booths is an open question. As far as we could tell, this was the only voting
precinct where any signs were posted about assisted voting, leaving us to wonder
whether those signs were having the possibly unanticipated consequence of
discouraging elderly voters from engaging in assisted voting.
9
Similarly, the Election Reform Information Project received multiple reports in
the 2004 general election of poll workers not understanding when or how to issue
provisional ballots (Cobb & Hedges, 2004). These cases are not unique, something
that led one election official to note that Poll workers are the Achilles heel of the
elections process and another to note that it seems remarkable that more
problems do not occur (Lush, 2004). The potential for error increases when the
law changes. For example, changes in the provisional voting law in Colorado led
poll workers to not give out provisional ballots to qualied voters, according to
several interest groups who monitor elections (Crist, 2004).
Furthermore, poll worker errors can have dramatic consequences. For
example, in the 2004 primary election in California, poll workers in 12 consoli-
dated precincts (voters from two distinct voting districts voting in one poll site)
in Orange County made a series of mistakes that resulted in voters receiving the
wrong ballot containing the wrong candidates and wrong voting districts. Thus,
in each consolidated poll site one of the voting precincts had voter turnout that
ranged from 5 percent to 21 percent, while the other voting precincts had
turnout totals ranging from 66 percent to 290 percent. As the Los Angeles Times
reported, poll workers unfamiliar with the new electronic voting system made
mistakes Tuesday that allowed many people to vote in the wrong districts,
potentially endangering the outcomes of several races (Rabin, Pfeifer, & Perry,
2004).
With the management of poll sites delegated to civic-minded individuals, the
training of poll workers and the monitoring of poll site operations becomes critical
to effective election administration. The GAO (2001a) estimates that 87 percent of
jurisdictions nationally provide some level of poll worker training. This training
varies from locality to locality, but typically contains information about the basics
of running the electionsfrom setting up the poll site to closing it downand
how to handle key problems, like when a voter comes to a poll site and cannot be
found on the registration roster. Many jurisdictions have created comprehensive
how-to guides for poll workers that explain step by step how to solve typical
problems. In order to encourage people to attend poll worker training, many LEOs
pay poll workers an additional stipend for attending. However, pre-election poll
worker training is not a panacea; for example, the poll workers who had trouble
implementing the new electronic voting system in Orange County, California in
the March 2004 primary had received training on the new system but still made
errors resulting in thousands of voters receiving the wrong ballots. Additionally, a
study of poll workers in Los Angeles County found that there was wide variation
in their level of understanding of basic election laws and procedures: Voters in
Alvarez/Hall: Controlling Democracy 497
poor and minority precincts were more likely to encounter a misinformed poll
worker than were their compatriots voting in a majority white precinct (Barreto,
Marks, & Woods, 2004).
Given the potential for poll workers to make errors, the obvious solution is to
increase monitoring efforts of poll site activities. Considering that election officials
have a difficult time nding qualied poll workers, it is not surprising that election
officials do not have the personnel to evaluate poll site operations effectively. As
the earlier GAO quote noted, few jurisdictions collect systematic data on polling
place operations. Instead, election officials rely on citizens with complaints or
interest groups that have an interest in monitoring poll activities to identify
problems. Third-party monitoring can provide election administrators with infor-
mation for identifying election problems, but there are obvious biases in the
precinct coverage they give and possibly in the information they provide (e.g.,
Hall, 2003 or Bjornlund, 2005).
Monitoring elections is complicated because it is not an ongoing activity in the
traditional manner of a government program or service. Elections occur on a single
day, perhaps three or four times per year. Therefore, election officials are typically
not in a position to collect longitudinal data or to rectify problems in a timely
manner, nor is there a political will to spend the necessary money to pay for a
professional approach to monitoring and managing elections. If a problem is
subtle, election officials might not nd out about it until the polls close, by which
time the problem can only be addressed for the next election. Even if the problem
is readily noticeable, it can take hours to remedy, during which time hundreds of
voters can be disenfranchised. Monitoring may be important for evaluating
election administration but in many cases the data can only be used in the next
election.
Polling Places
Adverse selection problems are not limited to poll workers; poll sites can be
similarly problematic. Problems at polling places represent both a resource
problemthere are often a limited number of potential polling places in a given
voting precinctas well as a delegation problem. Many states have laws that
govern the number of poll sites a jurisdiction must have, most commonly by
placing a cap on the number of voters who can vote at any one poll site. This
forces election officials to nd poll sites in specic areas in order to serve a given
population, and these sites may not be optimal. For example, the GAO (2001b)
found that 84 percent of poll sites used in the 2000 election were not fully
accessible to people with disabilities. The number of institutions willing to allow
their facilities to be used as a polling site is declining, even among public schools,
which have security and logistical concerns about their facilities being used as
polling sites (GAO, 2001a, pp. 165167). This problem exists despite many states
having laws requiring these institutions to provide facilities for poll sites upon
request of an election official.
498 Policy Studies Journal, 34:4
Even when election officials do nd appropriate sites, they still do not control
these facilities. They must rely on building managers or custodians to unlock the
buildings and ready the space for Election Day. Because the polls typically open
so early . . . custodians may not have opened the space so that the poll workers
could enter on time (GAO, 2001a, p. 167). Despite the best intentions, sometimes
polling places are inadequate for the task at hand; many lack electricity or
sufficient parking for voters. Some of the available locations, especially businesses
or schools, lack necessary requirements for polling places, especially during the
course of the business or school day, for example adequate and accessible parking
close to the polling place, or a facility that is free from distractions or disruptions.
10
Additionally, the owners of poll sites do cancel on election officials, often at the last
minute, leaving them scrambling for new sites (Hall, 2002).
As the previous GAO descriptions illustrate, principalagent theory is quite
applicable to an analysis of poll site voting on Election Day. In this analysis, the
election administrator is the principal working with a large number of agents
throughout the election process. On Election Day, the delegation of authority from
election officials to other people and entities is almost total. Most election admin-
istration is not conducted by professional election administrators. Instead, these
officials delegate control of the election to poll workers who are at best skilled
amateurs in election administration, and to the owner of poll sites, who often have
interests and preferences at odds with the election official.
11
Of course, this analysis
is not meant to diminish the well-meaning nature of everyone involved in the
election process, but instead to show the problems this system can entail. Delega-
tion brings to the fore the potential for adverse selection problems. Moreover, the
nature of elections and the spacing of elections in time make the tools that have
traditionally been used to constrain the activities of agents almost useless to an
election administrator.
Principalagent problems at poll sites are quite serious and inherent in the
process of election management. Even if problems do not occur at a given poll in
a given jurisdiction in a given election, the threat of these problems always exists.
For example, the host of a poll site can decide right before an election to use their
facility for a purpose other than voting, forcing the LEO to search for a new voting
location and to inform voters about the change in venue. Poll workers forget that
it is Election Dayas what happened in Los Angeles in 2002or willfully decide
not to show up for work, leaving poll sites unmanned or understaffed. These and
other problems can easily leave voters without a place to vote and disenfranchised,
even if only temporarily.
Alternatives to the Polling Place: Expanded Early Voting and Vote-by-Mail
Only in the smallest of jurisdictions can an election administrator single-
handedly overcome the principalagent problems inherent in poll site voting. Even
then, the problem can only be overcome if the administrator actually runs the poll
site. In any jurisdiction with more than a few poll sites, the question remains: How
can LEOs overcome these potential problems if they cannot run the election
Alvarez/Hall: Controlling Democracy 499
without delegating operations to agents? One answer is to minimize reliance on
poll sites and allow voters to vote from wherever they want in a given period of
time before an election. There are three current models for addressing this
problem: (i) making all voters absentee voters in a vote-by-mail (VBM) system, (ii)
creating large consolidated precincts, or vote centers, (iii) by utilizing expanded
in-precinct early voting. In the future, Internet voting could also serve to minimize
several major principal-agent problems. The advantage of each of these methods
is that each does not require the total delegation of election management to poll
workers and poll site owners, allowing the LEO to retain more control over the
election process. We discuss below how each of these systems addresses key
principalagent problems.
All-Voting-by-Mail
By moving the election process out of traditional polling places, there are no
longer polling places or poll workers to whom the administration of the election
has to be delegated. No voting machines, only vote tabulators, must be procured.
With vote-by-mail, the LEO only has to produce an absentee ballotan activity
easily under their direct controlprovide the ballot and election materials to
voters, and have ballot tabulation software that can count the ballots, a technology
that they need in any event. The only medium between the voter and the LEO is
a transit mediumtypically the post officealthough voters can return their ballot
to a physical location as well.
An important feature of VBM is that it has a track record, both in the United
States and abroad. Oregon was the rst state to use VBM as a means of
overcoming principalagent problems. In 1981, the state enacted a law allowing all
VBM trials to be conducted in local elections; in 1987, VBM could be used in all
special and local elections (Southwell & Burchett, 1997). Local governments
quickly adopted VBM because of the cost savings and the control they gained over
elections by not having to delegate election management to poll workers and poll
sites. By 1994, between 20 percent and 25 percent of all Oregonians were voting
absentee in the non-VBM statewide elections. In 1995, the Oregon Secretary of
State took advantage of U.S. senator Robert Packwoods resignation and ordered
that the elections to replace Packwood be carried out using VBM. Primary election
ballots were sent to the approximately 1.8 million registered voters. Voters could
either mail in the ballot or drop it off at designated sites, such as the county
courthouse. In the primary election, turnout was 58 percent of eligible voters; in
the general election, it was 66 percent, close to average for an Oregon congres-
sional election.
Since 1996, all statewide elections in Oregon have been held using VBM. A
1996 survey of Oregonians illustrates the ease of the VBM process (Southwell &
Burchett, 1997). In the 1995 VBM election, less than 2 percent of voters encountered
any difficulties in returning their ballot, in part because almost 90 percent of voters
cast their ballots within two weeks of receiving it (more than half voted in the rst
week). The survey also found that VBM enfranchised voters who traditionally
500 Policy Studies Journal, 34:4
have a difficult time voting at the polls. These included the approximately 15
percent of voters who traditionally have difficulty getting to the polls because of
physical disabilities, a lack of transportation, or time constraints inherent in their
job. Also, a signicantly higher percentage of non-white voters and single-parent
households voted, as did those who typically nd voting difficult because they
have moved recently or because they work. A more recent survey (Southwell,
2004) found that young people, housewives, and people with disabilities all
reported that VBM allowed them to vote more often.
Subsequent studies have identied how VBM affects other management
outcomes, including voter participation and the number of voter errors on the
ballot (Alvarez & Hall, 2004). In Oregon, data show that VBM may have a positive
effect on voting behavior. It encourages voters to participate more than any factor
other than it being a presidential election (Southwell & Burchett, 2000) and helps
retain voters in the voting electorate; voters that use VBM are likely to continue
voting in subsequent elections (Berinsky, Burns, Michael, & Traugott, 2001).
Because VBM captures and holds voters in the electorate, it has the effect of
increasing turnout.
12
VBM also reduces voter error rates. The Oregon Secretary of
State has calculated the ballot error rate in every county in every general election
since 1992.
13
The data show that, in the majority of Oregon counties that use optical
scan technology, the error rate has steadily declined in every presidential election
since 1992.
VBM has also been used in the United Kingdom. There, the Electoral Com-
mission has been encouraging local election administrators to try new innovations,
including VBM, which could improve voter turnout and improve election admin-
istration.
14
For example, a VBM pilot was conducted in 2001 in an election for the
West Wiltshire District Council, a local election where 257 voters were eligible to
vote. Voters had two weeks to cast their ballot and could return the ballot either
by mail or by dropping off the ballot at a designated collection point.
The local government conducted a voter education effort before the election,
informing all potential voters that the upcoming election would be conducted
using VBM. The result was quite positive, with the 56 percent turnout in the
special election being almost three times higher than in recent by-elections, and
almost 70 percent higher than turnout in the most recent general local election. It
was also cost-effective, with the LEO saving one-third of the typical cost for a
special election. The success of this and other VBM experiments has led the
Electoral Commission to call for all local elections to be run using VBM.
15
The costs
of sending ballots out by mail and educating voters about VBM are much lower
than the costs associated with actually running an election at the poll sites, where
poll workers must be paid, equipment purchased, and materials for poll workers
developed.
Early Voting
In addition to unrestricted absentee voting and VBM, early voting has also
changed the election process over the past two decades. Early voting is currently
Alvarez/Hall: Controlling Democracy 501
used in at least eight states and is especially prevalent in Texas, which has used
this voting method since 1988. Early voting frees voters from having to fulll their
civic duty within a 12-hour window on a Tuesday, allowing them to cast their
ballots at a poll site at their own convenience. It is also considered a more secure
voting method compared to the absentee voting process because it is conducted at
an official polling place, among observers and election officials, providing a more
secure, coercion-free, and private voting environment.
16
Early voting not only
makes voting easier, but also does not create partisan or major demographic
divides. Early voters are just as likely as Election Day voters to be Democrats or
Republicans, and are better-informed voters, in part because they tend to be strong
partisans. Early voting also is more likely to be used by the elderly and lower-
income voters, which is somewhat surprising since lower-income citizens are
generally considered to be low-propensity voters (e.g., Stein, 1998; Verba,
Schlozman, & Brady, 1995).
In 2004, early voting grew in popularity in many states across the nation. Over
10 million early votes were cast in the 2004 presidential election, accounting for 8.4
percent of ballots cast (EAC, 2005). In a number of states, early votes were a very
large fraction of total ballots cast: Texas, Tennessee, Nevada and Arizona all
reported that more than 4 of 10 ballots cast in 2004 were early votes (EAC). And
in two of these states (Texas and Tennessee), the proportion of ballots cast as early
votes has nearly tripled in the decade ending in 2004 (EAC).
At this point, there have been few studies of early voting in 2004; one
comprehensive study looked at early voting in Florida and found that early voting
created some administrative problems for election officials and did vastly stimulate
turnout by low-propensity voters (Gronke, Bishin, Stevens, & Galanes-Rosenbaum,
2005). What early voting does do for election officials is provide them with the
opportunity to serve many votersespecially special needs votersin facilities
that the LEO can better control and staff, with election workers who are more
experienced, and under the supervision of professional election staff. In addition,
by having more voters cast ballots before Election Day, the pressure on precinct
poll workers is reduced by lowering the volume of voters they have to handle.
Vote Centers
One new strategy for mitigating the problems with polling places on Election
Day is voting centers. The model jurisdiction in the implementation of vote centers
is Larimer County, Colorado. In Larimer County, a person can vote anywhere in
the county because the local election officials uses electronic poll books in their
polling places. These electronic poll books are connected to a secure central server
that updates voter registration data in real time, which keeps voters from voting
at multiple voting centers.
The twist is that Larimer County radically downsized the number of polling
places from 143 locations to just 22. These 22 polling places are in centrally located
cites that have large parking facilities and are very accessible to the disabled. In
many ways, the county stole a page from the blig box retailers, who have fewer
502 Policy Studies Journal, 34:4
mega locations but that are centrally located so they draw large numbers of
voters. The results were also successful; Larimer County reported higher turnout
after the switch. The vote centers allow election officials to have greater control
over the polling places that are used and use fewer poll workers. The reduction in
the number of poll workers needed allowed the county to cull out the poll workers
who were less effective but needed when there were more polling locations. Also,
with 22 locations, the local election official is in a position to exert more direct
management supervision over these locations.
However, to date there have been so few efforts to experiment with vote
centers in the United States that their ultimate impact on either election admin-
istration or voters is unknown (Stein, Leighley, & Owens, 2005). Clearly, vote
centers, if properly implemented, could ease the principal-agency problems asso-
ciated with widespread use of traditional polling places and volunteer poll
workers. But what costs might be associated with voting centers, and whether they
have any positive effect on voter participation, balloting mistakes by voters, voter
satisfaction, or any other outcome measure awaits future research.
Internet Voting
In the future, the Internet or other electronic technologies may become the
medium through which individuals register to vote and cast their ballot.
17
As with
VBM, registration and voting over the Internet or with other electronic technolo-
gies will allow election administrators to avoid delegation as much as possible
(Alvarez & Hall, 2004). Election administrators will be able to provide election
services directly to the voterincluding voter registration services and voting
servicesand voters may be able to use technologies like e-mail as a simple way
of communicating with election administrators. Again, the problems at poll sites
with poll workers and polling equipment will disappear. With Internet voting, the
LEO can control the voting equipment and how voters interact with election
services, although what the ultimate costs associated with the use of new tech-
nologies like the Internet for registration and voting also awaits future research.
The Difculty of Alleviating Agency Problems
VBM, early voting, vote centers, and in the future, new technologies like
Internet voting, may improve voting opportunities for voters and may provide
local election officials with greater control over the election process. However,
some critics have expressed concerns about whether these voting methods
especially VBM and Internet votingraise important privacy and fraud issues.
Even the administration of VBM can be problematic; in 2002 King County,
Washington, election officials had a breakdown in the absentee voting process that
may have disenfranchised thousandsillustrating that no system is completely
foolproof.
18
One important concern with vote centers is social science research that nds
that reducing the number of polling places can reduce turnout. For example, Brady
Alvarez/Hall: Controlling Democracy 503
and McNulty (2004) conducted a study examining the effects of consolidating
precincts in Los Angeles County for the 2003 recall election. They found that the
consolidation reduced turnout by 1.88 percent and also resulted in an increase in
absentee voting. Consolidation results in transportation effectsit is harder to get
to the pollsand a disruption effect from having to nd the new polling place.
Other scholars have found the greater the distance to the polls increases nonvot-
ing, and that the cost of nding less-accessible polling places is too high. These
results hold true even after controlling for motivational, informational, and
resource variables, and are quite pronounced in suburban precincts (Dyck &
Gimpel, 2005; Gimpel & Schuknecht, 2003).
With VBM and Internet voting, there is no guarantee that voters can mark and
seal their ballot in private, without coercion or pressure, even though survey data
suggest this may not be a problem (Southwell & Burchett, 1997). A traditional
polling place allows voters to stand in a voting booth or behind a curtain, alone,
after verifying their identity for a polling place worker. In addition, state laws are
quite strict about the partisan and political activities allowed in the vicinity of a
polling place. Concerns about the privacy of mailed ballots are especially acute for
voters in certain environments, such as nursing homes or other care facilities.
Privacy concerns also have been raised regarding overseas ballots, especially as
some military personnel have been allowed to fax their ballots to election officials.
Additionally, election offices can only verify the identity of a VBM voter by
comparing the signature on the ballot envelope with the signature on le. With
precinct voting and early voting, the voter is required to appear in person.
Although these privacy and fraud concerns are real, little evidence supports
claims of widespread coercion and lack of privacy in the VBM process. Specic
instances of absentee voting fraud have occurred: In a widely publicized case in
Miami in 1997, signicant fraud was alleged in the absentee voting process and the
courts eventually decided not to count any absentee ballots in determining the
election outcome.
19
One systematic study of Oregons vote-by-mail experience
noted only a handful of instances of vote-by-mail fraud in the United States, and
no demonstrable evidence of vote-by-mail fraud in OR despite that states exclu-
sive use of vote-by-mail (Gronke, 2005). Earlier studies of absentee voting also
found very little evidence of absentee voting fraud (Miller, 1948).
In the specic case of Internet voting, an array of concerns about the potential
for fraud has been raised (Alvarez & Hall, 2004; California Internet Voting Task
Force, 2000; NSF Internet Voting Workshop, 2001). For example, allowing citizens
to vote from their personal computer raises concerns about the security of each
individual computer platform used to vote, as well as the security of the ballot
transmission path and the security of the computer systems used by election
officials to receive, store, and tabulate electronic votes. These security concerns are
valid, although the odds of these security threats occurring are unknown and it is
unclear whether these threats are more signicant than those facing voting
methods currently used. Only by further research and development can the
potential effects of Internet-based registration and voting systems on election
administration problems be assessed (Alvarez and Hall).
504 Policy Studies Journal, 34:4
Critics have also argued that liberalized VBM and early voting affect the
nations civic values. These critics focus on the ceremonial and symbolic aspects of
Election Day votingsuch as everyone voting with their neighbors at a single
placeand how this fosters a sense of citizenship (Stromer-Galley, 2003). However,
there is no research indicating that, compared with precinct voters, absentee voters
have diminished civic values or that civic values in OR or CA are different than
in states like NY, which has restrictive absentee voting laws. Additionally, critics
tend to ignore the impact of long lines and other polling place problems on a
citizens civic values, not to mention the immediate impact that polling place
problems have on voting behavior.
20
Another related critique is that early voters do not cast informed votes. For
example, scholar Norman Ornstein wrote in 2001 that early voters voted from a
smaller base of knowledge than the rest of us had; it was the equivalent of
deciding the winner of a basketball game at the end of the third quarter.
21
Although early voters miss out on the last week or so of the campaigns, there is
no research indicating that absentee voters are less informed than precinct voters.
Additionally, many VBM voters in OR miss little of the election before they cast
their ballots. In the 2000 OR general election, election officials received 45 percent
of the VBM ballots in the nal two days of the election; in the 2002 general
election, election officials received 48 percent in the nal two days.
22
Implications
Some readers might question the application of principalagent theory to
election administration. After all, elections have historically been run by
amateurspolitical party members or individuals who are motivated by a sense
of civic-mindedness. However, as Robert Putnam has documented, there has been
a marked decline in civic engagement and in social capital over the last 50 years.
For example, membership in the League of Women Voters is down more than 60
percent from its peak in 1965 (Putnam, 2000, p. 439). Similarly, memberships in the
PTA and in religious organizations have declined as well, and as commuting times
increase and the household mobility rate remains at approximately 18 percent
annually, social capital will continue to decline. As Putnam (p. 346) notes, . . . the
performance of representative government is facilitated by the social infrastructure
of civic communities . . . In the language of economics, social capital lowers
transaction costs and eases dilemmas of collective action. It is in this world of
declining social capital that todays election officials operate.
Voting is the foundation of our democracy, yet across the nation the imple-
mentation of elections is done by amateurs who have little experiencein the most
extreme case they have some on-the-job training spread out over several years.
Major elections in the United States are held on a single day across the nation,
requiring LEOs to gear up in a massive effort to serve their entire community in
a single 12-hour time frame. The principalagent model illustrates that the ability
of the LEO to control the quality of service that is provided to their customerthe
voting publicis greatly limited by their inability to carefully select poll workers
Alvarez/Hall: Controlling Democracy 505
and to control the activities at poll sites. Even worse, if problems do arise, elections
are not an ongoing process in the typical sense. It is not as though the election
administrator is in a position to come in the day after the election and work
through the problems with the election workers as they do their jobs, as there may
not be another election for months, and the same individuals may not staff it. In
every election, the LEO has to make a new set of delegations to poll sites, poll
workers, and other parties.
Interestingly, there may be political incentives for LEOs to maintain the status
quo and keep the principalagent problems. The inefficiencies, the lack of formal
control over poll workers, and the ongoing likelihood for problems can take some
pressure off of the LEO to be responsible for the election. When things do go
wrong, the LEO can point to the current system and in part blame any of the facets
noted previouslythe poll workers, the vendors, the polling place ownersfor
the result. The lack of transparency and monitoring also can benet the LEO
because there are no systematic data to use to evaluate management performance
in the election. Even more serious, in places where local election officials are
elected or are openly partisan, the principalagent problems may not manifest
themselves randomly but instead be biased against voters or groups in a way that
benet the LEOs own political party.
There are ways to avoid or minimize delegation problems, but they require a
reconceptualization of elections outside of the traditional one-day, poll site extrava-
ganza used in America for more than two hundred years. There is an obvious
tension between managing and controlling elections and maintaining the historic
participatory democratic election process that involves polling places, poll
workers, and taking time out of ones day to engage in a civic ritual. However,
there is no reason why LEOs cannot recapture control of the election process by
removing the need to delegate election authority to amateur officials and still
maintain a feeling of civic engagement. Models existsuch as vote-by-mail, early
voting, vote centers, and Internet votingthat will allow election officials to
manage the voting process directly and the rise in the number of voters who are
seeking out these mechanisms suggest that there is quite a demand for them. Local
election officials should consider experimenting with these techniques to deter-
mine if this form of elections provides them with the ability to serve the public in
a more effective and efficient manner and to study the potential problems with
these alternative electoral procedures in a rational and scientic manner. Election
officials today often refer to multi-channel elections, where they offer voters
options to vote early, vote absentee, and also vote at polling places. They recognize
that the rst two options lessen pressure on poll sites on Election Day by
decreasing the volume of poll site voters, and also increase the level of public
service they are offering.
Thad E. Hall is assistant professor of Political Science and an Investigator with the
Institute for Public and International Affairs at the University of Utah.
R. Michael Alvarez is professor of Political Science at the California Institute of
Technology and is co-director of the Caltech/MIT Voting Technology Project.
506 Policy Studies Journal, 34:4
Notes
The authors thank Los Angeles County Registrar/Recorder Conny McCormack and Larimer County
Clerk Scott Doyle for their input, the small but growing cadre of election scholars for their comments,
and Mary King Sikora, Karen Kerbs and Melissa Slemin for their assistance.
1. In a recent example, a study by the United States General Accounting Office (GAO) found many
instances where election administrators were not adequately resourced for the complex tasks they
are required to perform (GAO, 2006). To note one specic example, the GAO estimated that 11
percent (of election jurisdictions) had an insufficient workforce for processing voter registration
applications (GAO, p. 85). This echoes the problems that Harris (1934) discussed, more than seven
decades earlier.
2. The report Voting: What Is, What Could Be, by the Caltech/MIT Voting Technology Project (VTP)
(2001) was the only report that examined elections beyond the scope of the current system,
although they too typically focused on improving the processes associated with poll site voting.
3. Examples include the works of Thompson (1998) and Brehm and Gates (1997). Additionally,
Frederich (1940) and Finer (1941) refer explicitly to principals and agents in their famous exchange
regarding accountability. Frederichs argument is a nonformalized statement of the principalagent
problem, with his explicit concern about informational asymmetries and conicts of interest
between principals and agents. Finer has similar concerns, albeit somewhat less explicitly argued,
that he thinks can only be solved if principals can use negative incentives of correction and
punishment, including allowing the principal to re the agent with relative ease.
4. Summaries of principalagent theory as applied in political science can be found in Bendor (1988),
Kiewiet and McCubbins (1991) (especially Chapter 2), Mitnick (1992), and Moe (1984).
5. The problems that do arise most typically involve language minority voters. For examples, see
GAO (1997); Justice Department Announces Resolutions in Two Florida Voting Matters, Depart-
ment of Justice, Press Release, #380: 0627-02, June 28, 2002; and About Language Minority Voting
Rights: Enforcement, U.S. Department of Justice, Civil Rights Division, Voting Section, http://
www.usdoj.gov/crt/voting/sec_203/activ_203.htm. But other problems arise, for example, regard-
ing ballot design; the buttery ballot used in Palm Beach County in the 2000 presidential election
is a prominent example (Montjoy & Slaton, 2002; Wand et al., 2001).
6. The most comprehensive source of information about the administration of elections in America is
the report of the U.S. GAO (2001a), in October 2001.
7. In a review of the 2002 March primary election in Los Angeles County, the report noted that 1,925
of the poll workers who attended training (from a total of 15,788 workers) did not end up serving
on Election Day. The report also noted that other counties in California were having trouble with
Precinct Inspector no shows on Election Day, with a 2 percent no show rate in Los Angeles
County on Election Day and a 3 percent rate in San Diego County. See McCormack (2002).
8. The polling place was number 90-0380A, located in the Solheim Lutheran Church, 2236 Merton
Avenue, Los Angeles.
9. Section 14282 and 14283 of Californias election code deal with assisted voting. In particular, Section
14282(a) states, When a voter declares under oath, administered by a member of the precinct board
at the time the voter appears at the polling place to vote, that the voter is then unable to mark a
ballot, the voter shall receive the assistance of not more than two persons selected by the voter . . .
As the voter is required to take this oath (and the polling place workers must keep a list of voters
who have needed assistance [Section 14283]), this implies that voters need to seek some permission
from the polling place workers if they need assistance with casting their ballot.
10. In recent election monitoring in Southern California, our team has observed a variety of such
problems with the use of private facilities for polling places, for example, polling in business
locations like tire shops or party stores, which have accessibility problems or which are noisy and
distracting locations to vote. Problems with the use of schools that are in-session abound in our
observation efforts, especially a lack of parking, problems with accessibility, and disruptions during
the school day as school children and teachers engage in their normal activities.
11. The problems associated with conicting goals among organizations is discussed in Rainey (1997),
Chapter 5.
Alvarez/Hall: Controlling Democracy 507
12. However, VBM does not tend to pull new votersincluding registered voters who have not
traditionally votedinto the electorate. If voters do not use VBM when initially introduced to it,
they will persist in ignoring the future ballots that are delivered as well.
13. Statistical Examination of Ballot Types in Oregon General Elections: 19922000. This report can
be accessed online at http://www.sos.state.or.us/executive/policy-initiatives/vbm/pcstudy.PDF.
14. The following discussion is taken from Pilot Scheme Evaluation, West Wiltshire District Council,
13 June 2002, The Electoral Commission, August 2002. The report is available online at http://
www.electoralcommission.gov.uk/publications_pdfs/pilot_reports/West%20Wiltshire.pdf.
15. Electoral Commission, Delivering Democracy? The Future of Postal Voting (August: London, UK,
2004).
16. The Caltech/MIT report (Caltech/MIT Voting Technology Project, 2001) ( July 2001) recommended
that liberalized absentee voting procedures be replaced with expanded early-voting opportunities.
It states, We have no systematic measures of fraud, but fraud appears to be especially difficult to
regulate in absentee systems. In-precinct voting or kiosk voting is observable. Absentee voting is
not. The prospect for coercion is increased with absentee voting on demand (p. 41).
17. While registration and voting over the Internet is the most commonly discussed application of
electronic technologies for early or absentee voting, other electronic technologies like e-mail, fax,
and text messaging have been used on trial bases in both the United States and the United
Kingdom.
18. See Secretary of State releases review of King County elections, NEWS RELEASE, February 13,
2003 for a summary of the issues and a link to the report. http://www.secstate.wa.gov/office/
news.aspx?news_id=192.
19. Judge Orders New Miami Mayoral Election, CNN, March 4, 1998 http://www.cnn.com/
ALLPOLITICS/1998/03/04/miami.mayor [July 22, 2003].
20. The Caltech/MIT Voting Technology Project (2001) estimated that of the four to six million votes
lost in the 2000 presidential election, as many as one million were lost because of polling place
problems.
21. The Dangers of Voting outside the Booth, New York Times, August 3, 2001 http://www.
aeipoliticalcorner.org/NO%20Articles/no010803.pdf [ July 22, 2003].
22. The 2000 data are available from the office of the Oregon secretary of state at http://www.sos.
state.or.us/elections/nov72000/other.info/g00byday.htm [July 22, 2003]; the 2002 data at http://
www.sos.state.or.us/elections/nov52002/g02byday.pdf [July 22, 2003].
References
Akerlof, George A. 1970. The Market For Lemons: Quality Uncertainty and the Market Mechanism.
Quarterly Journal of Economics 84 (3): 488500.
Alvarez, R. Michael, and Thad E. Hall. 2004. Point, Click, and Vote: The Future of Internet Voting.
Washington, DC: Brookings Institution Press.
. 2005. Rational and Pluralistic Approaches to HAVA Implementation: The Cases of Georgia and
California. Publius 35 (4): 55978.
Banks, Jeffrey S. 1989. Agency Budgets, Cost Information, and Auditing. American Journal of Political
Science 33 (3): 67099.
Banks, Jeffrey S., and Barry Weingast. 1992. The Political Control of Bureaucracies under Asymmetric
Information. American Journal of Political Science 36 (2): 50924.
Barreto, Matt A., Mara Marks, and Nathan Woods. 2004. Precinct Quality and Voter Turnout: Race,
Income, and Voter Participation. Presented at the Midwest Political Science Association Meeting,
Chicago, IL, April 15.
Bendor, Jonathan, Serge Taylor, and Roland van Gaalen. 1987. Politicians, Bureaucrats, and Asym-
metrical Information. American Journal of Political Science 31: 796828.
Bendor, Jonathan. 1988. Review Article: Formal Models of Bureaucracy. British Journal of Political
Science 18 (3): 35395.
508 Policy Studies Journal, 34:4
Berinsky, Adam J., Nancy Burns, and Michael W. Traugott. 2001. Who Votes by Mail? A Dynamic
Model of the Individual-level Consequences of Voting-by-Mail Systems. Public Opinion Quarterly
65: 17897.
Bjornlund, Eric C. 2005. Beyond Free and Fair: Monitoring Elections and Building Democracy. Baltimore,
MD: The Johns Hopkins University Press.
Brady, Henry, and John E. McNulty. 2004 The Costs of Voting: Evidence from a Natural Experiment.
Annual Meeting of the Society for Political Methodology, Stanford University, July 2931.
Brehm, John, and Scott Gates. 1997. Working, Shirking, and Sabatage: Bureaucratic Response to a Democratic
Public. Ann Arbor, MI: The University of Michigan Press.
California Interview Voting Task Force Report, http://www.ss.ca.gov/executive/ivote, Accessed:
January 18, 2000.
Caltech/MIT Voting Technology Project. 2001. Voting: What Is, What Could Be. http://vote.caltech.edu/
2001report.htm. Accessed October 9, 2006.
Cobb, Kim, and Michael Hedges. 2004. Large Voter Turnout Jams Polls. Houston Chronicle (November
3): A1.
Cooper, Phillip J. 2003. Government by Contracts. Washington, DC: CQ Press.
Crist, Gabrielle. 2004. Voter Rules Challenged. Rocky Mountain News (October 1): A26.
Donahue, John D. 1989. The Privatization Decision: Public Ends, Private Means. New York: Basic Books.
Dyck, Joshua J., and James G. Gimpel. 2005. Distance, Turnout, and the Convenience of Voting. Social
Science Quarterly 86 (3): 53148.
EAC 2005. Election Assistance Commission. Election Day Survey. http://www.eac.gov/election_
survey_2004/intro.htm Accessed: October 9, 2006.
Finer, Herman. 1941. Administrative Responsibility in Democratic Government. Public Administration
Review 1 (Summer): 33550.
Friedrich, Carl J. 1940. Public Policy and the Nature of Administrative Responsibility. Public Policy
324.
General Accounting Office. 1997. Bilingual Voting Assistance Provided and Costs. GAO/GGD-97-81.
. 2001a. Elections Perspectives on Activities and Challenges Across the Nation. GAO-02-3.
. 2001b. Voting with Disabilities: Access to Polling Places and Alternative Voting Methods.
GAO-02-107.
. 2006. Elections: The Nations Evolving Election System as Reected in the November 2004
General Election. GAO-06-450.
Gimpel, James, and Jason E. Schuknecht. 2003. Political Participation and the Accessibility of the Ballot
Box. Political Geography 22 (4): 47188.
Governors Select Task Force on Elections Procedures, Standards and Technology (Florida). 2001.
Revitalizing Democracy in Florida. The Collins Center for Public Policy, http://www.collinscenter.
org/USR.doc/RentalizingDemocracyinFlorida.pdf. Accessed October 9, 2006.
Gronke, Paul, Benjamin Bishin, Daniel Stevens, and Eva Galanes-Rosenbaum. 2005. Early Voting
in Florida, 2004. Presented at the 2005 Annual Meetings of the American Political Science
Association.
Gronke, Paul. 2005. Ballot Integrity and Voting By Mail: The Oregon Experience. Report prepared for
the National Commission on Election Reform [Online]. http://www.sos.state.or.us/executive/
CarterBaker.pdf. Accessed October 9, 2006.
Hall, Thad E. 2002. LA Story: The 2001 Election. A Century Foundation Report. New York: The Century
Foundation.
. 2003. Public Participation in Election Management: The Case of Language Minority Voters.
American Review of Public Administration 33 (4): 40722.
Hall, Thad E., and Laurence J. OToole, Jr. 2000. Structures for Policy Implementation: An Analysis of
National Legislation, 19651966 and 19931994. Administration and. Society 31 (6): 66786.
Alvarez/Hall: Controlling Democracy 509
Harris, Joseph P. 1934. Election Administration in the United States. Washington, D.C: Brookings
Institution.
Internet Policy Institute, Report of the National Workshop on Internet Voting: Issues and Research
Agenda, Washington, DC, March 2001.
Kiewiet, D. Roderick, and Mathew D. McCubbins. 1991. The Logic of Delegation. Chicago: The University
of Chicago Press.
Liebschutz, Sarah F., and Daniel J. Palazzolo. 2005. HAVA and the States. Publius 35 (4): 497514.
Lipksy, Michael. 1980. Street-Level Bureaucracy. New York: Russell Sage Foundation.
Lush, Tamara. 2004. Election Officials Worry Staff will be Achilles Heel at Polls. St. Petersburg Times
(August 7): B1.
McCormack, Conny B. 2002. March 5, 2002 Primary Election: Problems, Solutions and Resources
Needed for Improvement. Los Angeles County Registrar-Recorder/County Clerk
Miller, George Frederick. 1948. Absentee Voters and Suffrage Laws, Washington: Daylion Co., 1948.
Mitnick, Barry M. 1980. The Political Economy of Regulation. New York: Columbia University Press.
. 1992. The Theory of Agency and Organizational Analysis. In Ethics and Agency Theory: An
Introduction, ed. Norman E. Bowie, and R. Edward Freeman. New York: Oxford University Press,
Moe, Terry M. 1984. The New Economics of Organization. American Journal of Political Science 78:
73977.
Montjoy, Robert S., and Christa Daryl Slaton. 2002. Interdependence and Ethics in Election Systems:
The Case of the Buttery Ballot. Public Integrity 4 (Summer): 195210.
National Commission on Federal Election Reform. 2002. To Assure Pride and Condence in the Electoral
Process. Washington, DC: Brookings Institution Press.
Putnam, Robert D. 2000. Bowling Alone: The Collapse and Revival of American Community. New York:
Simon & Schuster.
Rabin, Jeffrey L., Stuart Pfeifer, and Tony Perry. 2004. Glitches Hinder Casting of Votes. Los Angeles
Times (March 3): B1.
Rainey, Hal G. 1997. Understanding & Managing Public Organizations, 2nd edn. San Francisco: Jossey-
Bass Publishers.
Southwell, Priscilla L. 2004. Five Years Later: A Re-Assessment of Oregons Vote by Mail Electoral
Process. PS, Political Science and Politics (January): 8993.
Southwell, Priscilla L., and Justin Burchett. 1997. Survey of Vote-by-Mail Senate Election in the State
of Oregon. PS, Political Science and Politics (March): 537.
. 2000. The Effect of All-Mail Eelections on Voter Turnout. American Politics Quarterly 28 (1):
729.
Stein, Robert. 1998. Early Voting. Public Opinion Quarterly 62: 5770.
Stein, Robert M., Jan Leighley, and Christopher Owens. 2005. Who Votes, Who Doesnt, and What Can
Be Done. A Report to the Federal Commission on Election Reform.
Stromer-Galley, Jennifer. 2003. Voting and the Public Sphere: Conversations on Internet Voting. PS,
Political Science and. Politics 36 (4): 72731. http://www.american.edu/iq/cfer/0630test.stein.pdf.
Accessed October 9, 2006.
Thompson, Fred. 1998. Public Economics and Public Administration. In Handbook of Public Admin-
istration, 2nd edn, ed. Hildreth Rabin, and Miller New York: Marcel Dekker, 9951063.
Verba, Sidney, Kay Schlozman and Henry Brady. 1995. Voice and Equality: Civic Voluntarism in American
Politics. Cambridge: Harvard University Press.
Wand, Jonathan N., Kenneth W. Shotts, Jasjeet S. Sekhon, Walter R. Mebane, Michael Jr., C. Herron, and
Henry E. Brady. 2001. The Buttery Did It: The Aberrant Vote for Buchanan in Palm Beach
County, Florida. American Political Science Review 47: 4660.
Waterman, Richard W., and Kenneth J. Meier. 1998. Principal-Agent Models: An Expansion? Journal
of Public Administration Research and Theory 8 (2): 173202.
510 Policy Studies Journal, 34:4